Microorganisms are the drivers of flavor compound production in cheese， and their community composition， dynamic succession， and resulting metabolites all influence the flavor and quality of cheese. In this study， metagenomics and metabolomics techniques were employed to analyze the microbial diversity and metabolite composition of traditional cheese samples from Ordos， Inner Mongolia. The aim was to decipher the correlation between microbial communities and the formation of cheese metabolites. The results revealed the presence of 8 phylum， 98 genus， and 205 species in the traditional cheese from Ordos. The cheese samples could be divided into two groups by principal component analysis. In ERDS-1 group， the dominant bacteria species were Lactobacillus helveticus and Lactococcus lactis. The main dominant bacteria species of ERDS-2 group cheese were Delbrueckii and Streptococcus thermophilus. The microbial diversity of cheese in ERDS-1 group was higher than that in ERDS-2 group， and the metabolic pathway glycolysis IV (PWY-1042) was significantly different between the two groups (P<0.05). This may be due to the differences in the production process， production temperature and microorganisms of the milk source， which lead to the difference in the structure of cheese flora， resulting in differences in metabolites. A total of 744 metabolites belonging to 14 classes were detected in the cheese samples， and 26 metabolites with significant differences were identified. Significant correlations were observed between certain bacterial species， such as Lactobacillus delbrueckii and Streptococcus thermophilus， and metabolites content such as calcitriol， UDP-glucose， and lyso-phosphatidylcholine in the cheese. The results of this experiment provide a theoretical basis for the mining of core bacteria and key metabolites in cheese， and provide a reference for the industrial production of traditional cheese.